US7937175B2 - Methods and systems for pervasive diagnostics - Google Patents

Methods and systems for pervasive diagnostics Download PDF

Info

Publication number
US7937175B2
US7937175B2 US12/170,542 US17054208A US7937175B2 US 7937175 B2 US7937175 B2 US 7937175B2 US 17054208 A US17054208 A US 17054208A US 7937175 B2 US7937175 B2 US 7937175B2
Authority
US
United States
Prior art keywords
plant
production
plan
plans
diagnostic
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active, expires
Application number
US12/170,542
Other languages
English (en)
Other versions
US20100010654A1 (en
Inventor
Johan de Kleer
Robert Price
Lukas D. Kuhn
Rong Zhou
Minh Do
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xerox Corp
Original Assignee
Palo Alto Research Center Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Palo Alto Research Center Inc filed Critical Palo Alto Research Center Inc
Priority to US12/170,542 priority Critical patent/US7937175B2/en
Assigned to PALO ALTO RESEARCH CENTER INCORPORATED reassignment PALO ALTO RESEARCH CENTER INCORPORATED ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: DE KLEER, JOHAN, DO, MINH, KUHN, LUKAS D., PRICE, ROBERT, ZHOU, RONG
Priority to EP09164975.6A priority patent/EP2144129B1/en
Priority to JP2009163423A priority patent/JP5107974B2/ja
Publication of US20100010654A1 publication Critical patent/US20100010654A1/en
Application granted granted Critical
Publication of US7937175B2 publication Critical patent/US7937175B2/en
Assigned to XEROX CORPORATION reassignment XEROX CORPORATION ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: PALO ALTO RESEARCH CENTER INCORPORATED
Assigned to CITIBANK, N.A., AS COLLATERAL AGENT reassignment CITIBANK, N.A., AS COLLATERAL AGENT SECURITY INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: XEROX CORPORATION
Assigned to XEROX CORPORATION reassignment XEROX CORPORATION CORRECTIVE ASSIGNMENT TO CORRECT THE REMOVAL OF US PATENTS 9356603, 10026651, 10626048 AND INCLUSION OF US PATENT 7167871 PREVIOUSLY RECORDED ON REEL 064038 FRAME 0001. ASSIGNOR(S) HEREBY CONFIRMS THE ASSIGNMENT. Assignors: PALO ALTO RESEARCH CENTER INCORPORATED
Assigned to JEFFERIES FINANCE LLC, AS COLLATERAL AGENT reassignment JEFFERIES FINANCE LLC, AS COLLATERAL AGENT SECURITY INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: XEROX CORPORATION
Assigned to CITIBANK, N.A., AS COLLATERAL AGENT reassignment CITIBANK, N.A., AS COLLATERAL AGENT SECURITY INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: XEROX CORPORATION
Assigned to XEROX CORPORATION reassignment XEROX CORPORATION TERMINATION AND RELEASE OF SECURITY INTEREST IN PATENTS RECORDED AT RF 064760/0389 Assignors: CITIBANK, N.A., AS COLLATERAL AGENT
Active legal-status Critical Current
Adjusted expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
    • G05B19/41865—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM] characterised by job scheduling, process planning, material flow
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00—Testing or monitoring of control systems or parts thereof
    • G05B23/02—Electric testing or monitoring
    • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0208—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the configuration of the monitoring system
    • G05B23/021—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the configuration of the monitoring system adopting a different treatment of each operating region or a different mode of the monitored system, e.g. transient modes; different operating configurations of monitored system
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00—Testing or monitoring of control systems or parts thereof
    • G05B23/02—Electric testing or monitoring
    • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
    • G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
    • G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/32—Operator till task planning
    • G05B2219/32083—Alternative, variant operation planning, revision specification of product
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/30—Nc systems
    • G05B2219/33—Director till display
    • G05B2219/33299—Real time, online diagnostic, integrated in normal control system
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/80—Management or planning

Definitions

  • the present exemplary embodiments relate to automated diagnosis and production in systems having multiple resources for achieving production goals.
  • automated diagnosis of system performance and component status can advantageously aid in improving productivity, identifying faulty or underperforming resources, scheduling repair or maintenance, etc.
  • Accurate diagnostics requires information about the true condition of components in the production system. Such information can be obtained directly from sensors associated with individual components and/or may be inferred from a limited number of sensor readings within the production plant using a model or other knowledge of the system structure and dynamics. Providing complete sensor coverage for all possible system faults can be expensive or impractical in harsh production environments, and thus it is generally preferable to instead employ diagnostic procedures to infer the source of faults detected or suspected from limited sensors.
  • System diagnostic information is typically gathered by one of two methods, including dedicated or explicit diagnostics with the system being exercised while holding production to perform tests and record observations without attaining any production, as well as passive diagnostics in which information is gathered from the system sensors during normal production. Although the latter technique allows inference of some information without disrupting production, the regular production mode may not sufficiently exercise the system to provide adequate diagnostic information to improve long term productivity. Moreover, while dedicated diagnostic operation generally provides better information than passive diagnostics, the cost of this information is high in terms of short term productivity reduction, particularly when diagnosing recurring intermittent system component failures that require repeated diagnostic interventions.
  • the present disclosure provides systems and methods for implementing on-line pervasive diagnostics in a production system, in which a planner constructs/generates a plan in the form of a sequence of actions taken by one or more resources in a plant based on production goals and at least partially based on one or both of production and diagnostic objectives or metrics.
  • Goals as used herein include binary conditions that either are or are not achieved, wherein production goals as used herein include the production of products, modification of products, etc.
  • Model-based production control systems and production plan construction/generation methods are provided for controlling operation of a production system, in which a planner constructs plans for execution in the plant based on production goals and diagnostic objectives, and a diagnosis engine determines a current plant condition based on a previously executed plan and corresponding observations from the plant, and provides expected information gain data to the planner.
  • Objectives as used herein constitute aspirations toward which systems are directed, and may be achieved in whole or in part.
  • production objectives may include, for example, production of products at a higher rate, or of higher quality, or more efficiently, etc.
  • diagnostic objectives may include, for instance, isolating the source of system faults, etc.
  • the planner constructs a plan that will achieve a given production goal based at least partially on a diagnostic objective and the expected information gain data.
  • this novel approach provides significant advantages over conventional passive diagnostics in the amount of information that may be obtained, and is significantly more cost effective in terms of short term productivity and may also improve long run production objectives compared with conventional dedicated/explicit diagnostic techniques.
  • a model-based production control system including a model, a planner, and a diagnosis engine for controlling the operation of a production system that has one or more plants with resources to achieve production goals.
  • the plant model includes a model of the plant and the planner provides one or more plans for execution in the plant based on at least one output objective.
  • the diagnosis engine determines a current plant condition based on one or more previously executed plans, at least one corresponding observation from the plant, and the plant model.
  • the diagnosis engine also provides expected information gain data to the planner for one or more possible plans based on the current plant condition and the model.
  • the planner generates or constructs a plan that will achieve a given production goal at least partially based on a diagnostic objective and the expected information gain data.
  • This control system facilitates the inclusion of diagnostic objectives and expected diagnostic information gain into the generation of production plans, and thus may be implemented to provide more useful diagnostic information than conventional passive diagnostic techniques, while still avoiding or mitigating system down-time.
  • the control system is flexible in allowing the planner to also construct or generate dedicated diagnostic plans based on one or more particular diagnostic objectives and the current plant condition, where the information to be gained may substantially improve the ability to realize long term productivity improvements.
  • the planner may selectively interleave dedicated diagnostic plans and production plans for execution in the plant based on at least one production goal and at least one diagnostic objective.
  • the diagnostic objective may be updated based at least partially on the expected information gain data.
  • the control system may also include an operator interface that allows an operator to provide observations that the diagnosis engine can use in determining the current plant condition and a diagnosis job description language allowing definition of a diagnostic plan.
  • the current plant condition and the plant model may be used at least partially by the planner to generate/construct a control plan or a sequence of control plans for the plant that will achieve a set of production goals while balancing both production objectives (e.g., production cost, production time) and diagnostic objectives (e.g., diagnostic cost, information gained, repair cost) to perform diagnosis in isolating faulty resources in the plant.
  • production objectives e.g., production cost, production time
  • diagnostic objectives e.g., diagnostic cost, information gained, repair cost
  • a current plant condition is determined based at least partially on a previously executed plan, at least one corresponding observation from the plant, and a plant model.
  • the method also includes determining expected information gain data for one or more possible plans based on the current plant condition and the model, and constructing a plan that will achieve a given production goal based at least partially on a diagnostic objective and the expected information gain data.
  • the plan construction may also be based on the current plant condition, and may include making a tradeoff between production objectives and diagnostic objectives and/or may further include performing diagnosis to isolate faulty resources in the plant at least partially based on the current plant condition.
  • the method may further include constructing a dedicated or explicit diagnostic plan as well as selectively interleaving dedicated diagnostic plans and production plans based on at least one production goal and at least one diagnostic objective.
  • the method may include allowing a user to define a diagnostic plan using a diagnosis job description language, as well as receiving user observations and constructing the plan at least partially based on the user observations.
  • a computer readable medium which has computer executable instructions for performing the steps of determining a current plant condition based at least partially on a previously executed plan, at least one corresponding observation from the plant, and a plant model, and determining expected information gain data for one or more possible plans based on the current plant condition and the model.
  • Computer executable instructions are also provided for constructing a plan for execution in the plant that will achieve a given production goal based at least partially on a diagnostic objective and the expected information gain data.
  • the computer readable medium may also include computer executable instructions for constructing the plan based at least partially on the current plant condition.
  • FIG. 1 is a schematic diagram illustrating a production system and an exemplary model-based control system with a planner, a plant model, a diagnosis engine, and a operator interface in accordance with one or more aspects of the disclosure;
  • FIG. 2 is a schematic diagram illustrating further details of an exemplary modular printing system plant in the production system of FIG. 1 ;
  • FIG. 3 is a schematic diagram illustrating further details of the exemplary planner and diagnosis engine in the control system of FIGS. 1 and 2 ;
  • FIG. 4 is a schematic diagram illustrating a plan space for a production system, including production and diagnostic plans
  • FIG. 5 is a flow diagram illustrating an exemplary method for constructing plans for execution in a production system in accordance with one or more aspects of the present disclosure
  • FIG. 6 is a flow diagram illustrating an exemplary method of evaluating and generating plans for execution in the plant using an A* search
  • FIG. 7 is a schematic diagram illustrating an exemplary state/action diagram showing possible plans for transitioning the system state from a starting state to a goal state;
  • FIGS. 8-16 are graphs showing simulated cost vs. time curves for passive, explicit, and pervasive diagnosis techniques in the system of FIGS. 1-3 ;
  • FIG. 17 is a schematic flow diagram illustrating construction of plans in the system of FIGS. 1-3 using a SAT solver in accordance with further aspects of the disclosure.
  • the concepts of the disclosure also find utility in association with product packaging systems and any other type or form of system in which a plurality of resources, whether machines, humans, software or logic components, objects, etc., may be selectively employed according to plans comprised of a series of actions to achieve one or more production goals based at least partially on one or more diagnostic metrics or objectives, wherein all such alternative or variant implementations are contemplated as falling within the scope of the present disclosure and the appended claims.
  • the disclosure finds particular utility in constructing and scheduling plans in systems in which a given production goal can be achieved in two or more different ways, including use of different resources (e.g., two or more print engines that can each perform a given desired printing action, two different substrate routing paths that can be employed to transport a given printed substrate from one system location to another, etc.), and/or the operation of a given system resource at different operating parameter values (e.g., operating substrate feeding components at different speeds, operating print engines at different voltages, temperatures, speeds, etc.).
  • resources e.g., two or more print engines that can each perform a given desired printing action, two different substrate routing paths that can be employed to transport a given printed substrate from one system location to another, etc.
  • different operating parameter values e.g., operating substrate feeding components at different speeds, operating print engines at different voltages, temperatures, speeds, etc.
  • FIGS. 1-3 illustrate on such system 1 in which the various aspects of the present disclosure may be implemented.
  • a production system 6 is illustrated including a producer component 10 that receives production jobs 49 from a customer 4 and a plant 20 having a plurality of resources 21 - 24 that may be actuated or operated according to one or more plans 54 so as to produce one or more products 52 for provision to the customer 4 by the producer 10 , where ‘producing’ products can include modifying products, objects, etc., including without limitation packaging or wrapping products.
  • FIG. 2 illustrates further details of one exemplary plant 20 and FIG. 3 shows additional details regarding the exemplary model-based control system 2 .
  • the producer 10 manages one or more plants 20 which actually produce the output products 52 to satisfy customer jobs 49 .
  • the producer 10 in this embodiment provides jobs and objectives 51 to a multi-objective planner 30 of the model-based control system 2 and the production system 6 receives plans 54 from the planner 30 for execution in the plant 20 .
  • the jobs 54 can include one or both of production and diagnostic goals.
  • the control system 2 further includes a plant model 50 with a model of the plant 20 , and a diagnosis engine 40 with a belief model 42 .
  • the diagnosis engine 40 determines and updates a current plant condition 58 via a plant condition estimation/updating component 44 ( FIG. 3 ) based on one or more previously executed plans 54 , corresponding observations 56 from the plant 20 , and the model 50 .
  • the diagnosis engine 40 also provides expected information gain data 70 to the planner 30 for one or more possible plans 54 based on the current plant condition 58 and the model 50 .
  • the model-based control system 2 and the components thereof may be implemented as hardware, software, firmware, programmable logic, or combinations thereof, and may be implemented in unitary or distributed fashion.
  • the planner 30 , the diagnosis engine 40 , and the model 50 are software components and may be implemented as a set of sub-components or objects including computer executable instructions and computer readable data executing on one or more hardware platforms such as one or more computers including one or more processors, data stores, memory, etc.
  • the components 30 , 40 , and 50 and sub components thereof may be executed on the same computer or in distributed fashion in two or more processing components that are operatively coupled with one another to provide the functionality and operation described herein.
  • the producer 10 may be implemented in any suitable hardware, software, firmware, logic, or combinations thereof, in a single system component or in distributed fashion in multiple interoperable components.
  • the control system 2 may be implemented using modular software components (e.g., model 50 , planner 30 , diagnosis engine 40 and/or sub-components thereof) to facilitate ease of debugging and testing, the ability to plug state of the art modules into any role, and distribution of operation over multiple servers, computers, hardware components, etc.
  • FIG. 1 also includes an optional operator interface 8 implemented in the computer or other platform(s) on which the other components of the control system 2 are implemented, although not a strict requirement of the disclosure, wherein the operator interface 8 may alternatively be a separate system operatively coupled with the control system 2 .
  • the exemplary operator interface 8 is operatively coupled with the diagnosis engine 40 to provide operator observations 56 a to the diagnosis engine 40 , with the diagnosis engine 40 determining the current plant condition 58 based at least partially on the operator observations 56 a in certain implementations.
  • the exemplary operator interface 8 allows the operator to define a diagnostic job 8 b using a diagnosis job description language 8 a , and the diagnosis engine 40 may provide diagnostic jobs 60 to the producer 10 .
  • the diagnosis engine 40 in this implementation is operative to selectively provide one or more self-generated diagnostic jobs 60 and/or operator defined diagnostic jobs 8 b to the producer 10 , which in turn provides jobs and objectives 51 to the planner 30 .
  • the planner 30 provides one or more plans 54 to the production system 6 for execution in the plant 20 based on at least one output objective 34 ( FIG. 3 ) and production goals as directed by the incoming jobs 51 from the producer 10 .
  • the planner 30 selectively factors in one or more output objectives 34 derived from the jobs and objectives 51 in constructing plans 54 including production objectives 34 a and diagnostic objectives 34 b .
  • the production objectives 34 a are created and updated according to the jobs and objectives 51 obtained from the production system 6
  • the diagnostic objectives 34 b are derived from and updated according to the current plant condition 58 and the expected information gain data 70 provided by the diagnosis engine 40 .
  • the production objectives 34 a in one implementation may relate to the scheduling of orders for produced products 52 ( FIG. 1 ), and may include prioritization of production, minimization of inventory, and other considerations and constraints driven in large part by cost and customer needs.
  • Examples of production objectives 34 a include prioritizing plan construction/generation with respect to achieving a given product output goal (simple production criteria) as well as a secondary consideration such as simple time efficient production, cost efficient production, and robust production.
  • cost efficient production objectives 34 a will lead to construction/generation of plans 54 that are the most cost efficient among the plans that met the production goal as dictated by the jobs 51 received from the producer 10 .
  • the diagnostic objectives 34 b may include objectives related to determining preferred action sequences in generated plans 54 for performing a given production-related task, minimization of maintenance and repair costs in operation of the plant 20 , identifying resources 21 - 24 causing intermittent or persistent faults, etc.
  • the control system 2 may optionally include a plan data store or database 36 used to store plans 54 selectable by the planner 30 for execution in the plant 20 to facilitate one or more production or diagnostic objectives 34 , wherein construction/generation of a plan 54 as used herein can include selection of one or more pre-stored plans 54 from the data store 36 .
  • the planner 30 can selectively re-order a job queue so as to improve the likelihood of information gain.
  • the plan data store 36 may be provided in a separate component or components that are operatively coupled with the planner 30 by which the planner 30 can obtain one or more plans 54 (whole and/or partial) therefrom.
  • the planner 30 can synthesize (e.g. construct or generate) one or more plans 54 as needed, using the plant model 50 and information from the producer 10 and diagnosis engine 40 to determine the states and actions required to facilitate a given production and/or diagnostic objectives 34 .
  • the planner 30 creates and provides plans 54 for execution in the plant 20 .
  • the plans 54 include a series of actions to facilitate one or more production and/or diagnostic objectives 34 while achieving a production goal according to the jobs 51 , and in which a given action may appear more than once.
  • the actions are taken with respect to states and resources 21 - 24 defined in the model 50 of the plant 20 , for example, to route a given substrate through a modular printing system 20 from a starting state to a finished state as shown in FIG. 2 .
  • the planner 30 generates or constructs a plan 54 that will achieve a given production goal at least partially based on a diagnostic objective 34 b and the expected information gain data 70 from the diagnosis engine 40 .
  • the planner 30 in the illustrated embodiment includes a goal-based plan construction component 32 that assesses the current plant condition 58 from the diagnosis engine 40 in generating a plan 54 for execution in the plant 20 .
  • the component 32 may also facilitate identification of faulty components 21 - 24 or sets thereof in constructing the plans 54 based on observations 56 and current plant conditions 58 indicating one or more plant components 21 - 24 as being suspected of causing system faults.
  • the presently disclosed intelligent plan construction techniques advantageously provide for generation of plans 54 for execution in the plant 20 within a plan space 100 that includes both production plans 102 and diagnostic plans 104 .
  • the union of the plan sets 102 and 104 includes production plans 106 that have diagnostic value (e.g., can facilitate one or more diagnostic objectives 34 b in FIG. 3 ), wherein the planner 30 advantageously utilizes information from the diagnosis engine 40 to preferentially construct plans 106 that achieve production goals while obtaining useful diagnostic information in accordance with the diagnostic objectives 34 b .
  • the intelligent plan construction aspects of the present disclosure thus integrate the production planning and diagnosis to facilitate the acquisition of more useful diagnostic information compared with conventional passive diagnostic techniques without the down-time costs associated with conventional dedicated diagnostics.
  • the diagnostic information gained can be used to improve the long term productivity of the system 6 , thereby also facilitating one or more production objectives 34 a ( FIG. 3 ).
  • the diagnosis engine 40 in one embodiment includes a belief model 42 representing the current state of the plant 20 , and a component 44 that provides the current condition of the plant 20 to the planner 30 based on the previous plan(s) 54 and corresponding plant observations 56 .
  • the component 44 also estimates and updates the plant condition of the belief model 42 according to the plant observations 56 , the plant model 50 , and the previously executed plans 54 .
  • the operator observations 56 a from the interface 8 may also be used to supplement the estimation and updating of the current plant condition by the component 44 .
  • the estimation/updating component 44 provides the condition information 58 to inform the planner 30 of the confirmed or suspected condition of one or more resources 21 - 24 or other components of the plant 20 ( FIG. 1 ).
  • This condition information 58 may be considered by the plan construction component 32 , together with information about the plant 20 from the plant model 50 in providing plans 54 for implementing a given production job or goal 51 , in consideration of production objectives 34 a and diagnostic objectives 34 b .
  • the diagnosis engine 40 also includes a component 46 that provides expected information gain data 70 to the planner 30 based on the model 50 and the belief model 42 .
  • the information gain data 70 may optionally be determined in consideration of the operator defined diagnostic jobs 8 b from the operator interface 8 .
  • FIG. 2 illustrates further details of an exemplary modular printing system plant 20 in the production system 6 , including material supply component 21 that provides printable sheet substrates from one of two supply sources 21 a and 21 b , a plurality of print or marking engines 22 , an output finisher station 23 , a modular substrate transport system including a plurality of bidirectional substrate transport/router components 24 (depicted in dashed circles in FIG. 2 ), one or more output sensors 26 disposed between the transport system 24 and the finisher 23 , and a controller 28 providing control signals for operating the various actuator resources 21 - 24 of the plant 20 .
  • the exemplary printing system plant 20 includes four print engines 22 a , 22 b , 22 c , and 22 d , although any number of such marking engines may be included, and further provides a multi-path transport highway with three bidirectional substrate transport paths 25 a , 25 b , and 25 c , with the transport components 24 being operable by suitable routing signals from the controller 28 to transport individual substrate sheets from the supply 21 through one or more of the marking engines 22 (with or without inversion for duplex two-side printing), and ultimately to the output finishing station 23 where given print jobs are provided as output products 52 .
  • Each of the printing engines 22 may individually provide for local duplex routing and media inversion, and may be single color or multi-color printing engines operable via signals from the controller 28 .
  • the model-based control system 2 may, in certain embodiments, be integrated into the plant controller 28 , although not a strict requirement of the present disclosure.
  • the planner 30 automatically generates plans 54 , for example, by piece-wise determination of a series of actions to form a plan and/or by obtaining whole or partial plans 54 from the data store 36 for component resources 21 - 24 of the printing system plant 20 from a description of output production goals derived from the incoming jobs 51 in consideration of one or more production objectives 34 a and diagnostic objectives 34 b .
  • plans 54 for example, by piece-wise determination of a series of actions to form a plan and/or by obtaining whole or partial plans 54 from the data store 36 for component resources 21 - 24 of the printing system plant 20 from a description of output production goals derived from the incoming jobs 51 in consideration of one or more production objectives 34 a and diagnostic objectives 34 b .
  • the plant 20 has flexibility in how the output goals can be achieved (e.g.
  • the diagnosis engine 40 can alter or influence the plan construction operation of the planner 30 to generate a plan 54 that is expected to yield the most informative observations 56 .
  • the constructed plan 54 in this respect may or may not result in a plan that compromises short term production objectives 34 a (e.g., increases job time or slightly lowers quality), but production nevertheless need not be halted in order for the system to learn.
  • the additional information gained from execution of the constructed job 54 can be used by the producer 10 and/or by the planner 30 and diagnosis engine 40 to work around faulty component resources 21 - 24 , to schedule effective repair/maintenance, and/or to further diagnose the system state (e.g., to confirm or rule out certain system resources 21 - 24 as the source of faults previously detected by the sensor(s) 26 ).
  • the information gleaned from the constructed plans 54 e.g., plant observations 56
  • the estimation and updating component 44 can be used by the estimation and updating component 44 to further refine the accuracy of the current belief model 42 .
  • the plant 20 includes only limited sensing capabilities, (e.g., such as the system in FIG. 2 having only sensors 26 at the output of the transport system 24 downstream of the printing engines 22 ), passive diagnosis are unable to unambiguously identify every possible fault in the system 20 , whereas direct diagnostic efforts lead to system down-time and the associated cost in terms of productivity.
  • the control system 2 of the present disclosure advantageously facilitates selective employment of intelligent on-line diagnosis though construction and execution of plans 54 that provide enhanced diagnostic information according to the plant condition 58 and/or the expected information gain 70 , and may further advantageously facilitate generation of one or more dedicated diagnostic plans 54 for execution in the plant 20 based on at least one diagnostic objective 34 b and the plant condition 58 , and for intelligent interleaving of dedicated diagnostic plans 54 and production plans 54 based on production and diagnostic objectives 34 according to the current plant condition 58 .
  • the planner 30 can cause execution of explicit diagnostic plans 54 that involve halting production when the information gained from the plan 70 is expected to lead to significant future gains in productivity, enhanced ability to identify faulty resources 21 - 24 , or other long term productivity objectives 34 a and/or diagnostic objectives 34 b.
  • the control system 6 significantly expands the range of diagnosis that can be done online through pervasive diagnostic aspects of this disclosure during production (e.g., above and beyond the purely passive diagnostic capabilities of the system), thereby lowering the overall cost of diagnostic information by mitigating down time, the number of service visits, and the cost of unnecessarily replacing components 21 - 24 in the system 20 that are actually working, without requiring complete sensor coverage.
  • the planner 30 is further operative to use the current plant condition 58 in making a tradeoff between production objectives 34 a and diagnostic objectives 34 b in generating plans 54 for execution in the plant 20 , and may also take the condition 58 into account in performing diagnosis in isolating faulty resources 21 - 24 in the plant 20 .
  • the plant condition estimation and updating component 44 of the diagnosis engine 40 infers the condition of internal components 21 - 24 of the plant 20 at least partially from information in the form or observations 56 derived from the limited sensors 26 , wherein the diagnosis engine 40 constructs the plant condition 58 in one embodiment to indicate both the condition (e.g., normal, worn, broken) and the current operational state (e.g., on, off, occupied, empty, etc.) of the individual resources 21 - 24 or components of the plant 20 , and the belief model 42 can be updated accordingly to indicate confidence in the conditions and/or states of the resources or components 21 - 24 .
  • the diagnosis engine 40 receives a copy of the executed plan(s) 54 and corresponding observations 56 (along with any operator-entered observations 56 a ).
  • the condition estimation and updating component 44 uses the observations 56 , 56 a together with the plant model 50 to infer or estimate the condition 58 of internal components/resources 21 - 24 and updates the belief model 42 accordingly.
  • the inferred plant condition information 58 is used by the planner 30 to directly improve the productivity of the system 20 , such as by selectively constructing plans 54 that avoid using one or more resources/components 21 - 24 known (or believed with high probability) to be faulty, and/or the producer 10 may utilize the condition information 58 in scheduling jobs 51 to accomplish such avoidance of faulty resources 21 - 24 .
  • the exemplary diagnosis engine 40 also provides future prognostic information to update the diagnostic objectives 34 b which may be used by the planner 30 to spread utilization load over multiple redundant components 21 - 24 to create even wear or to facilitate other long term objectives 34 .
  • the diagnosis engine 40 provides the data 70 to the planner 30 regarding the expected information gain of various possible production plans 54 .
  • the planner 30 can use this data 70 to construct production plans 54 that are maximally diagnostic (e.g., most likely to yield information of highest diagnostic value).
  • the planner 30 can implement active diagnostics or active monitoring by using carefully generated or modified production plans 54 to increase information during production (e.g., using ‘diagnostic’ production plans).
  • certain diagnostic plans 54 are non-productive with respect to the plant 20 , but nevertheless may yield important diagnostic information (e.g., operating the transport mechanisms 24 in FIG.
  • the operator interface 8 allows an operator to create diagnostic jobs 8 b via the job description language 8 a
  • the diagnosis engine 40 may also include a diagnosis job description language to generate dedicated/explicit diagnostic jobs 60 which are provided to the producer 10 .
  • the producer 10 may then provide these jobs 60 to the planner 30 along with the other jobs and objectives 51 to explicitly request the planner 30 to advance diagnostic objectives 34 b .
  • the producer 10 in one implementation may operate a job queue that queues requested customer and diagnostic jobs 49 , 60 and the producer 10 receives component condition updates 58 from the diagnosis engine 40 .
  • the producer 10 uses the condition 58 to choose between customer jobs 49 and diagnosis jobs 60 , to tradeoff production efficiency versus diagnostic value in production plans 54 , and to merge (e.g., interleave) customer jobs 49 and dedicated diagnostic jobs 60 when they are compatible and wherein the scheduling thereof can facilitate one or more diagnostic and production objectives 34 .
  • the diagnosis engine 40 can also provide prognostic information to the planner 30 to help improve the quality of the plans 54 with respect to certain criteria.
  • the planner 30 (e.g., and/or the producer 10 ) is operative to selectively use fault state information to construct from multiple suitable production plans 54 based on the prognosis of plan alternatives for “robust printing” to distribute workload evenly across different resources 21 - 24 in order to reduce the frequency of scheduled or unscheduled maintenance of the plant 20 .
  • an exemplary method 200 is illustrated for constructing plans 54 for execution in a production system 6 with a plant 20 having a plurality of resources 21 - 24 to achieve one or more production goals. While the method 200 is illustrated and described below in the form of a series of acts or events, it will be appreciated that the various methods of the disclosure are not limited by the illustrated ordering of such acts or events. In this regard, except as specifically provided hereinafter, some acts or events may occur in different order and/or concurrently with other acts or events apart from those illustrated and described herein in accordance with the disclosure. It is further noted that not all illustrated steps may be required to implement a process or method in accordance with the present disclosure, and one or more such acts may be combined.
  • the illustrated method 200 may be implemented in hardware, software, or combinations thereof, such as in the exemplary control system 2 described above, and may be embodied in the form of computer executable instructions stored in a computer readable medium, such as in a memory operatively associated with the control system 2 in one example.
  • Diagnostic objectives 34 b are received at 202 in the method 200 .
  • the method 200 further includes determining a current plant condition 58 at 204 based at least partially on a previously executed plan 54 and at least one corresponding observation 56 from the plant 20 using a plant model 50 , and determining expected information gain data 70 at 206 based on the current plant condition 58 and the model 50 .
  • the planner 30 receives the plant conditions 58 at 208 from the diagnosis engine 40 , and receives production jobs and objectives 51 at 210 from the producer 10 .
  • the planner 30 constructs a plan 54 at based at least partially on a diagnostic objective 34 b and the expected information gain data 70 .
  • the planner 30 sends the constructed plan 54 to the plant 20 for execution and the diagnosis engine 40 receives the plan 54 and the plant observations 56 at 216 .
  • the diagnosis engine 40 updates the plant condition 58 and updates the expected information gain data 70 , after which further jobs and objectives 51 are serviced and the process 200 continues again at 210 as described above.
  • the plan construction at 212 may be based at least partially on the current plant condition 58 , and may include making a tradeoff between production objectives 34 a and diagnostic objectives 34 b based at least partially on the current plant condition 58 . Moreover, the plan construction at 212 may include performing prognosis to isolate faulty resources 21 - 24 in the plant 20 based at least partially on the current plant condition 58 .
  • a dedicated diagnostic plan 54 may be constructed for execution in the plant 20 based at least partially on at least one diagnostic objective 34 b , a diagnostic job 60 , 8 b , and the current plant condition 58 , and the plan construction may provide for selectively interleaving dedicated diagnostic and production plans 54 based on at least one production objective 34 a and at least one diagnostic objective 34 b . Further embodiments of the method 200 may also include allowing an operator to define a diagnostic plan 8 b using a diagnosis job description language 8 a and receiving operator observations 56 a , with the plan selection/generation at 216 being based at least partially on the operator observations 56 a.
  • a computer readable medium which has computer executable instructions for performing the steps of determining a current plant condition 58 in a production system 6 with a plant 20 having a plurality of resources 21 - 24 based at least partially on a previously executed plan 54 , at least one corresponding observation 56 from the plant 20 , and a plant model 50 , and computer executable instructions for determining expected information gain data 70 based on the current plant condition 58 and the model 50 .
  • the medium further includes instructions for constructing a plan 54 for execution in the plant 20 to achieve one or more production goals based at least partially on a diagnostic objective 34 b and the expected information gain data 70 .
  • further computer executable instructions are included in the medium for constructing the plan 54 based at least partially on the current plant condition 58 .
  • FIGS. 6 and 7 a variety of techniques can be employed by the planner 30 in the above described control system 6 in constructing plans 54 to enhance diagnostic information gain while achieving production goals.
  • a heuristic search is employed by the planner 30 in constructing diagnostic production plans 54 .
  • FIG. 6 illustrates an exemplary method 300 for evaluating and constructing plans for execution in the plant 20 using A* search.
  • the component 44 of the diagnosis engine 40 ( FIG. 3 ) establishes and updates beliefs in the belief model 42 about the plant conditions and updates the plant model 50 .
  • the component 44 of the diagnosis engine 40 FIG. 3
  • FIG. 3 establishes and updates beliefs in the belief model 42 about the plant conditions and updates the plant model 50 .
  • the component 44 derives failure or fault probabilities for each fault hypothesis in the plant 20 and employs dynamic programming at 304 to generate a upper and lower plan failure probability bounding heuristic based on the belief model 42 and the current plant condition 58 .
  • the planner 30 uses the heuristic to evaluate partial plans 54 so as to construct maximally informative plans 54 , preferably using an A* search approach.
  • the diagnosis engine 40 then uses the constructed plan 54 and output observations 56 obtained from execution of the constructed plan 54 in the plant 20 to update the belief model 42 .
  • the diagnosis engine 40 in this approach advantageously provides the inputs for searching by the planner 30 in order to derive valuable information for the diagnosis of the system 20 .
  • the best plans 54 with respect to diagnostic value for single persistent faults are those that have an equal probability of succeeding or failing.
  • the diagnosis engine 40 uses this notion to develop heuristics to guide the search by the planner 30 in evaluating partial plans 54 to construct the plan 54 to be executed in the plant 20 .
  • the plan construction search may employ pruning techniques to improve search performance.
  • control system 2 implements efficient on-line active or pervasive diagnosis in controlling the plant 20 through a combination of model-based probabilistic inference in the diagnosis engine 40 with decomposition of the information gain associated with executing a given plan 54 using an efficient heuristic target search in the planner 30 .
  • this active diagnosis technique specific inputs or control actions in the form of plans 54 are constructed by the planner 30 with the help of the diagnosis engine 40 to maximize or increase the amount and/or quality of diagnostic information obtained from the controlled system plant 20 .
  • the informative production (active or pervasive diagnosis) techniques of the present disclosure rank the set of partial plans at any given point in the construction process that achieve production goals by their potential information gain and the planner 30 operates to selectively generate the most promising strategy in that respect, in consideration of other possible production objectives 34 a and diagnosis objectives 34 b.
  • the planner 30 operates to construct the sequence of actions (plan 54 ) that transfers substrate sheets through the system 20 to generate a requested output for a given print job (e.g. to satisfy a production goal), using pervasive diagnosis to aid in plan construction.
  • One particular production objective 34 a in this system 20 is to continue printing even if some of the print engines 22 fail or some of the paper handling modules 24 fail or jam.
  • there are only output type sensors 26 preceding the finisher 23 and as a result, a plan 54 consisting of numerous actions must be executed before a useful observation 56 can be made.
  • the diagnostic engine 40 updates its belief model 42 and the current condition 58 to be consistent with the executed plan 54 and the observations 56 .
  • the diagnosis engine 40 forwards updated condition information 58 and the expected information gain data 70 to the planner 30 .
  • a search algorithm may be employed to find and construct plans 54 .
  • the model 50 describes the plant system 20 as a state machine with all possible actions A that the plant 20 can accommodate. Actions are defined by preconditions and post-conditions over the system state. As such, an action requires the system 20 to be in a certain state in order to be executable and modifies the system state when executed.
  • the system 20 is controlled by plan p ( 54 ) that is comprised of a sequence of actions a 1 , a 2 , . . . , a n drawn from the set A of possible actions.
  • Execution of an action potentially changes the system state, and part of the system state may represent the state of a product 52 at any given time, particularly if the action is part of a production plan 54 . Further, internal constraints of the system 54 limit the set of plans 54 to a subset of all possible sequences (e.g., the plan space 100 in FIG. 4 above). Moreover, the execution of actions of a given plan p in the system 20 may result in only a single observable plan outcome or observation O (e.g., observation 56 from sensor 26 ).
  • observable outcome 26 is defined as abnormal, denoted ab(p), in which the plan 54 fails to achieve its production goal.
  • Another type is a normal (not abnormal) outcome, denoted ab(p) in which the plan 54 achieves the associated production goal.
  • H SYS a set of all hypotheses excluding a single hypothesis (a “no fault” hypothesis h 0 ) for the situation where all actions are normal, every hypothesis is a complete assignment of abnormality to each action, and all are unique and mutually exclusive (e.g., ⁇ h i , h j ⁇ H sys , h i ⁇ h j ).
  • the system's beliefs in the belief model 42 can be represented as a probability distribution over the hypothesis space H sys , Pr(H), where the belief model 42 is updated by the diagnosis engine 40 from past observations 56 using Bayes' rule to get a posterior distribution over the unknown hypothesis H given observation O and plan P: Pr(H
  • O, P) ⁇ Pr(O
  • a plan p is deemed by the diagnosis engine 40 and the planner 30 as being informative if it contributes information to (e.g., reduces the uncertainty of) the diagnosis engine's beliefs 42 , where the informative value can be measured as the mutual information between the system beliefs Pr(H) and the plan outcome conditioned on the plan executed, I(H; O
  • P p).
  • This mutual information may be defined in terms of entropy or uncertainty implied by a probability distribution, where a uniform probability distribution has high uncertainty and a deterministic distribution has low uncertainty.
  • an informative plan 54 reduces the uncertainty of the system's beliefs 42 , and thus plans 54 with outcomes that are hard to predict are the most informative, while execution of plans 54 that are known to succeed (or known to fail) will yield no diagnostic information gain.
  • an optimal (e.g., ‘target’) uncertainty T may be used to rate the plans 54 with respect to expected informative value, by which the diagnosis engine 40 can evaluate plans 54 accordingly and provide the expected information gain data 70 to the planner 30 to influence the plan selection/generation for preferential selection/generation of informative plans 54 .
  • a plan 54 will be abnormal ab(p) if one or more of its actions are abnormal, as set forth in the following equation (1): ab(a 1 ) ⁇ . . . ⁇ ab(a n ) ab(p) for a i ⁇ A p (1) where a i ⁇ A p , and p is the plan.
  • the probability of a plan failure Pr(ab(p)) can be defined as the sum of all probabilities of hypotheses which imply that the plan will fail, as in the following equation (4):
  • the planner 30 evaluates the plans 54 in terms of the probability T and uses this and the current plant state 58 to preferentially construct a plan 54 that achieves a production goal while having a positive probability of failure. While in the short term this may likely lower the productivity of the system 20 , the information gained allows improvement in long term productivity.
  • the planner 30 moreover, may construct a sequence of plans 54 for execution in the plant 20 which might not be maximally informative individually, but are maximally informative taken jointly.
  • any form of search may be employed in the planner 30 that piece-wise selects/generates from suitable plans that will achieve a given production goal while yielding useful diagnostic information within the scope of the present disclosure.
  • the diagnosis engine 40 may establish a heuristic by which the planner 30 considers sets or families of plans 54 that share structure, such as by employing an A* target search using a set of partial plans p I ⁇ S 1 , p I ⁇ S 2 , . . . , p I ⁇ S n which progress from an initial state I to intermediate states S 1 , S 2 , . . . , S n .
  • the planner 30 uses the A* target search to attempt to expand the plan most likely to achieve the production goal in the best (e.g., most informative) way.
  • An ideal plan p in this regard would start with the prefix p I ⁇ S n which takes the system 20 to a state S n and continues with the suffix plan p S n ⁇ G leading from the state S n to the goal state G.
  • the planner 30 employs a heuristic function derived by the diagnosis engine 40 based at least partially on the description of the system architecture and dynamics in the plant model 50 .
  • FIG. 7 shows an exemplary state/action diagram 400 depicting possible plans for transitioning the system state from a starting state S to a goal state G in the plant 20 .
  • the system state nodes 402 include the starting state S 402 s , the goal stage G 402 g , and four intermediate states 402 a - 402 d for nodes A-D, respectively.
  • a given plan 54 for this example proceeds by following actions 404 through the diagram 400 to ultimately reach the goal G 402 g .
  • One possible plan 54 that satisfies such a production goal moves the system through the state sequence [S, A, C, G] through actions 404 sa , 404 ac , and 404 cg as shown in FIG. 7 .
  • this plan 54 results in an abnormal outcome caused by a faulty action 404 ac between nodes A and C (action a A,C ), caused by a single persistent fault in one of the system resources 21 - 24
  • the diagnosis engine 40 would determine from the plan 54 and the resulting fault observation 56 that all of the actions 404 sa , 404 ac , and 404 cg along the plan path are (without further information) suspected of being faulty. Since a single persistent fault is assumed, there are three positive probability hypotheses corresponding to the suspected actions ⁇ a S,A ⁇ , ⁇ a A,C ⁇ , ⁇ a C,G ⁇ . Absent additional information, the diagnosis engine 40 initially assigns equal probabilities ⁇ 1 ⁇ 3 ⁇ , ⁇ 1 ⁇ 3 ⁇ , ⁇ 1 ⁇ 3 ⁇ to these suspected actions.
  • the diagnosis engine 40 uses the graph structure and probability estimates to construct heuristic bounds on the uncertainty that can be contributed to a plan by any plan suffix, in this example, by building up the heuristic from right to left in FIG. 7 .
  • the exemplary diagnosis engine 40 assigns lower and upper bounds [L,U] to the nodes 402 , as shown in FIG. 7 , and these bound values are sent to the planner 30 in one implementation.
  • the action a D,G leading from state D to the goal state G in FIG. 3 was not part of the observed plan 54 that failed, and is therefore not a candidate hypothesis, and this action has a zero probability of being the source of the assumed single persistent system fault.
  • State B 402 b likewise has a lower bound of zero as plans 54 passing through state B can be completed by an action a B,D 404 bd that does not use a suspected action 404 and ends in state D which has a zero lower bound.
  • State B in this example has an upper bound of 1 ⁇ 3 since it can be completed by an unsuspected action a B,C 404 bc to state C 402 c which in turn has both upper and lower bounds with 1 ⁇ 3 probability of being abnormal.
  • the diagnosis engine 40 continues this analysis recursively to determine bounds on the probability of a suffix sub-plan being abnormal, and sends these as part of the information gain data 70 to the planner 30 .
  • the planner 30 uses these bounds with a forward A* target search to identify and construct a plan 54 that achieves or most closely approximates the target probability T.
  • one possible plan 54 begins from the start node S 402 s and includes a first action a S,A , which was part of the plan 54 that was observed to be abnormal. If the action a S,A 404 ac is added to a partial plan, it must add 1 ⁇ 3 probability to the chance of failure as it is a candidate itself. After a S,A , the system 20 would be in state A, and a plan 54 could be completed through D by including actions 404 ad and 404 dg to arrive at the goal state G 402 g .
  • the action a A,D itself has a zero probability of being abnormal since it was not involved in the previously observed faulty plan. Using the heuristic bound, therefore, a completion through state node D 402 d adds zero probability of being abnormal. From node A 402 a , a plan 54 could alternatively be completed through node C, as in the originally observed plan 54 . The corresponding action a A,C 404 ac adds 1 ⁇ 3 probability of failure to such a plan and based on the heuristic bound the completion through C 402 c must add another 1 ⁇ 3 probability of being abnormal.
  • the heuristic that is precomputed and provided by the diagnosis engine 40 therefore allows prediction of total plan abnormality probability for a possible plan 54 that moves the system 20 through the state node sequence [S, A, C, G] or [S, A, D, G].
  • the lower bound of the total plan is 1 ⁇ 3, as determined by 1 ⁇ 3 from a S,A plus 0 from the completion a A,D ,a D,G , and the upper bound is 3/3 equal to the sum of 1 ⁇ 3 from a S,A plus 1 ⁇ 3 each from a A,C and a C,G . If this plan is computed through [a A,C ,a C,G ] the total plan 54 will fail with probability 1, and therefore nothing is to be learned from constructing such a plan completion.
  • the planner 30 will construct the plan 54 [S, A, D, G] for execution in the plant 20 .
  • the plan 54 may or may not succeed, and in either case something may be learned from a diagnostic perspective. For instance, if the plan [S, A, D, G] fails, the diagnosis engine 40 learns that node a S,A was the failed action (for the assumed single persistent fault scenario), and if the plan 54 is successful, the engine 40 can further refine the belief model 42 by eliminating action 404 sa as a fault suspect.
  • a successor state S n of a state S m is any state that can be reached in a single step starting from the state S m .
  • the search for an informative production plan by the planner 30 starts from the initial starting state S 402 a and works recursively forward toward the goal state 402 g .
  • the abnormality probability of the empty plan starting at the initial state S is zero plus the best completion.
  • the planner 30 computes the abnormality probability as the plan probability up to the current state plus the abnormality probability of the best completion route. Since the planner 30 is initially uncertain about the completion, its probability of abnormality is an interval that includes a lower and upper bound and the values in between.
  • the total abnormality probability is also an interval, as set forth in the following equation (7):
  • I ( p I ⁇ Sn ) [ Pr ( ab ( p I ⁇ Sn ))+ L Sn ,Pr ( ab ( p I ⁇ Sn ))+ U Sn )] (7)
  • the planner 30 can therefore construct an interval describing how close the abnormality probabilities will be to T according to the following equation (8):
  • This absolute value in equation (8) folds the range around T, and if the estimated total abnormality probability of the plan 54 straddles target probability T, then the interval
  • the exemplary function F has some advantageous properties.
  • the planner 30 may be operative to improve the efficiency of the target plan search using selective pruning.
  • the above described search heuristic in many cases may return the same value, i.e., zero, which provides the planner 30 with little guidance in making a selection.
  • the planner 30 may accordingly be adapted to focus the search using one or more techniques.
  • the planner 30 prunes out dominated parts of the search space. For example, a given partial plan I(p I ⁇ S n ) may be identified by the planner 30 having an abnormality probability interval that does not straddle the target value T. The best possible plan in this interval will be on the one of the two boundaries of the interval that is closest to the target value T.
  • V pI ⁇ S n min(
  • the plan p I ⁇ S i will dominate every plan p I ⁇ S j where ⁇ T ⁇ I(p I ⁇ S ).
  • the planner 30 accordingly operates to prune out (eliminate from further consideration) some or preferably all dominated plans from the A* search space.
  • the planner 30 may employ other focusing techniques to intelligently break ties in the heuristic value.
  • the heuristic value determines which state node will be expanded next, but it is possible that two or more nodes will receive the same heuristic value.
  • the planner 30 may employ one or more rules to break the tie and hence to determine which node should be expanded first.
  • One suitable rule in this regard is to simply pick a node randomly.
  • V pI ⁇ S n (in equation (9) above) represents a guaranteed lower bound on a total plan P I ⁇ G starting with the partial plan p I ⁇ S n as prefix. While the upper and lower bounds are realizable, none of the interior points of the interval are guaranteed to exist. Therefore, the planner 30 may advantageously compare the V's in order to decide which of two partial plans has the closest realizable solution to break the above mentioned ties. Moreover, if two partial plans are also identical in this parameter, the information gain is the same, and thus the planner 30 is operative to choose the partial plan that has less likelihood to fail, thereby facilitating short term productivity.
  • the planner 30 can also facilitate the selective avoidance of known faulty resources 21 - 24 in the plant 20 via the component 32 b , as well as generation of plans 54 so as to help determine the source of faults observed during production.
  • the planner 30 operating the above described modular printing system plant 20 of FIG. 2 can be influenced by diagnostic objectives 34 b ( FIG. 3 ) to preferentially construct paper paths via appropriate routing of substrates to use different subsets of routing and printing components 24 and 22 , where a given sequence of these paths can be used to isolate the cause of an observed fault.
  • multiple plant pathways, redundancy of plant resources, and the capability to operate resources at different speeds, voltage levels, temperatures, or other flexibility in setting operational parameters of the plant resources allows the planner 30 to tailor active production plan generation for intelligent diagnostic information gain despite lack of complete sensor coverage in a given plant 20 .
  • the modularity and flexibility of a given system 20 can be exploited by the pervasive diagnostic features of the control system 2 to facilitate diagnostic objectives 34 b while also providing benefits with regard to flexibility in achieving production goals.
  • the control system 2 can thus provide the advantages of performing diagnosis functions during production, even with limited sensor capabilities, with the flexibility to schedule dedicated diagnostic plans 54 if/when needed or highly informative.
  • the planner 30 focuses on the needs of the diagnosis engine 40 and thus creates/selects plans 54 that maximize information gain with respect to the fault hypotheses.
  • the system 2 also allows the generation of plans 54 solely on the basis of production goals, for instance, where there is only one plan 54 that can perform a given production task and the planner 30 need not chose from a set of equivalent plans, thereby limiting the information gathering to the case of passive diagnosis for that plan.
  • the control system 2 can choose to parallelize production to the extent possible, use specialized print engines 22 for specific printing tasks, and have the operational control to reroute sheet substrates around failed modules as these are identified.
  • the planner 30 may receive a production print job 51 from a job queue (in the producer 10 , or a queue in the planner 30 ), and one or more plans 54 are constructed as described above to implement the job 51 .
  • the observations 56 are provided to the diagnosis engine 40 upon execution of the plan(s) 54 to indicate whether the plan 54 succeeded without faults (e.g., not abnormal), or whether an abnormal fault was observed (e.g., bent corners and/or wrinkles detected by the sensors 26 in printed substrates).
  • the diagnosis engine 30 updates the hypothesis probabilities of the belief model 42 based on the executed plan 54 and the observations 56 .
  • the planner 30 constructs the most informative plan 54 in subsequent scheduling so as to satisfy the diagnostic objectives 34 b .
  • there may be a delay between submitting a plan 54 to the plant 20 and receiving the observations 56 and the planner 30 may accordingly plan production jobs 51 from the job queue without optimizing for information gain until the outcome is returned in order to maintain high short term productivity in the plant 20 .
  • the plan construction in the planner 30 is biased to have an outcome probability closest to the target T, and this bias can create paths capable of isolating faults in specific actions.
  • the plant 20 Prior to detection of a system fault, the plant 20 may produce products 52 at a nominal rate r nom , with diagnosis efforts beginning once some abnormal outcome is observed.
  • the length of time required to diagnose a given fault in the system (e.g., to identify faulty plant components or resources 21 - 24 ) will be short if dedicated, explicit diagnostic plans 54 are selected, with pervasive diagnosis approaches taking somewhat longer, and passive diagnostic techniques taking much longer and possibly not being able to completely diagnose the problem(s).
  • diagnosis cost With regard to diagnosis cost, however, explicit dedicated diagnosis results in high production loss (production is halted), while purely passive diagnosis incurs the highest expected repair costs due to its lower quality diagnosis.
  • the pervasive diagnosis aspects of the present disclosure advantageously integrate diagnostic objectives 34 b into production planning by operation of the planner 30 , and therefore facilitate realization of a lower minimal total expected production loss in comparison to passive and explicit diagnosis.
  • the passive diagnostic aspects of the disclosure are generally applicable to a wide class of production manufacturing problems in which it is important to optimize efficiency but the cost of failure for any one job is low compared to stopping the production system to perform explicit diagnosis.
  • the disclosure finds utility in association with non-manufacturing production systems, for example, service industry organizations can employ the pervasive diagnostic techniques in systems that produce services using machines, software, and/or human resources.
  • the disclosure is not limited to a probability based A* search, wherein other plan construction techniques can be employed such as a SAT-solver approach in which the clauses represent failed plans and each satisfying assignment is interpreted as a valid diagnosis.
  • the above described heuristic search by the planner 30 was simulated to evaluate the cost advantages of pervasive diagnosis in a model of a modular digital printing press based on the above described printing plant 20 in FIG. 2 .
  • this modular system 20 multiple pathways allow parallel production, use of specialized print engines 22 for specific sheets, and selective rerouting around failed modules 21 , 22 , etc.
  • the planner 40 sends a plan 54 to a simulation of the printing system 20 which models the physical dynamics of the paper moving through the system 20 and determines the outcome.
  • plan 54 is completed without any faults (e.g., bent corners or wrinkles) and deposited in the requested finisher tray 23 , the plan 54 is deemed to have succeeded (not abnormal), and otherwise the plan 54 outcome is determined to be abnormal.
  • the original plan 54 and the observed outcome 56 are sent to the diagnosis engine 40 which updates the hypothesis probabilities of the belief model 42 .
  • a system module e.g., one of the print engines 22
  • the simulation of costs employed a simple cost model of opportunity costs in terms of unrealized production due to efforts of isolating the faulty component (diagnosis costs) and exchanging this component (repair costs).
  • the cost in this model represents the expected total amount of lost production due to a fault. When a fault is first observed, the cost is 1 (lost unit of production) and the belief state 42 is a uniform distribution over all fault hypothesis for all approaches.
  • the graph 510 in FIG. 8 shows instantaneous repair cost for the cases of passive diagnostics 511 , explicit (dedicated) diagnostics 512 and pervasive diagnostics 513 .
  • Cumulative diagnosis costs vs. time are shown in the graph 520 of FIG. 9 , including curves 521 , 522 , and 523 for passive, explicit, and pervasive diagnostic techniques.
  • FIGS. 11-13 depict instantaneous repair cost, cumulative diagnosis cost, and response cost graphs 610 , 620 , and 630 , respectively, as a function of time.
  • the graph 610 in FIG. 11 illustrates passive, explicit, and pervasive cost curves 611 , 612 , and 613 , respectively, for instantaneous repair cost.
  • cumulative diagnosis cost curves 621 , 622 , and 623 are shown for passive, explicit, and pervasive diagnostic techniques
  • Graph 710 in FIG. 14 shows passive, explicit, and pervasive instantaneous repair cost curves 711 , 712 , and 713 , respectively.
  • FIG. 15 shows curves 722 , 723 , and 723 for passive, explicit, and pervasive diagnosis, respectively.
  • the graph 730 shows passive, explicit, and pervasive diagnosis response cost curves 731 , 732 , and 733 , respectively.
  • the instantaneous repair costs in FIGS. 8 , 11 , and 14 was computed by estimating the repair time based on the current probability distribution over the fault hypothesis and pricing this downtime according to the nominal machine production rate.
  • the diagnosis cost in FIGS. 9 , 12 , and 15 at a given time t was calculated as the accumulated production deficit in relation to a healthy machine producing at its nominal rate r nom , where the x-axis is the amount of time (relative to the first occurrence of the fault) after which one chooses to stop diagnosis and start repairing the machine.
  • the minimum of the sum these costs denotes the optimal point in time (relative to first occurrence of the fault) to switch from diagnosis to repair and gives the minimal expected total loss of production due to the fault.
  • the optimal response cost of pervasive diagnosis is below those of the other two approaches, and the shortest diagnosis duration is for explicit, with pervasive and passive diagnosis techniques taking successively longer.
  • the explicit diagnostic approach involves the highest production loss since production is halted during diagnosis, whereby explicit diagnosis does not result in minimal response costs.
  • Passive diagnosis has the lowest rate of lost production, but incurs the highest expected repair costs due to its lower quality diagnosis, due to plans 54 being constructed and executed solely on the basis of production goals and objectives 34 a irrespective of diagnostic objectives 34 b .
  • Pervasive diagnosis intelligently integrates the diagnostic objectives 34 b into production planning by using the flexibility of the planner 30 and the diagnosis engine 40 , and the simulated results show a lower minimal total expected production loss in comparison to passive and explicit diagnosis approaches.
  • the planner 30 may employ a Boolean Satisfiability problem (SAT) solver 38 ( FIG. 3 ) in selecting plans 54 for execution in the plant 20 using the guidance of the diagnosis engine 40 .
  • SAT Satisfiability problem
  • the pervasive diagnosis tasks are translated into logical encodings (e.g., CNF, DNNF, BDD, PI, HTMS) to include all possible bounded length plans, and the encoding can be solved in the planner 30 using a SAT solver 38 to answer diagnosis queries such as finding a plan that uses a certain set of modules 22 in the plant 20 .
  • This technique allows the planner to submit queries to the SAT solver 38 instead of having to search for the plans 54 via A* heuristic search or simple brute force search techniques as described above.
  • FIG. 17 illustrates this approach in a diagram 800 .
  • the process begins with a CNF formulation that encapsulates all possible plans with fixed bound on (i.e., a sequence of bounded length executable actions by a given system). Additional variables corresponding to modules are included in the encoding. Additional formulas are included to ensure that a module variable is TRUE only if some action that uses that module is executed.
  • the plan construction component 32 formulates the plan generation problem based on the production goals and diagnostic objectives 34 b , and uses the current plant condition 58 from the diagnosis engine 40 to derive a list of suspected actions and/or plant modules 22 at 812 . These are provided to a translation component 804 in the planner 30 that generates a SAT formulation 814 .
  • the SAT problem formulation 814 is then provided to a SAT solver 38 , such as by a query to the solver component 38 in FIG. 3 , and the solver 38 generates a SAT solution 816 .
  • the solution 816 is then translated at 808 to yield a generated plan 54 at 818 for execution in the plant 20 .
  • a diagnostic objective 34 b may be to reduce the set of suspect modules 22 in the plant 20
  • the solver 38 generates a set of modules 22 that should be used in the next plan execution, for which a set of modules can be identified by the solver 38 using a maximizing entropy heuristic.
  • the plan construction component 32 queries the SAT solver 38 to construct a set of modules 22 having close to 0.5 probability of failure.
  • the generated plan 54 is identified by the solver 38 , it is provided to the plant 20 by the planner 30 and executed.
  • the plan 54 and the corresponding observations 56 are provided to the diagnosis engine 40 which updates the current condition 58 and the belief model 42 to reduce the set of suspected modules 22 , and the diagnosis engine 40 or the planner 30 add appropriate formulas to the SAT encoding to reflect those changes. This process is continued in one implementation until the set of suspected modules cannot be reduced.
  • the SAT solver approach in FIG. 17 is not as easily scaled up for large complex systems 20 , but is adaptable to simple modular systems, and may advantageously employ off-the shelf SAT solvers to solve the plan generation problem in the planner 30 .
  • the SAT solver approach captures all plans 54 , and a wide range of queries can be posted to the SAT solver 38 by the plan construction component 32 in generating a plan to accommodate diagnostic objectives 34 b .
  • the SAT problem can be formulated in any suitable encoding, including without limitation Constraint Satisfaction (CSP) and Integer Linear Programming (ILP).

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • General Engineering & Computer Science (AREA)
  • Manufacturing & Machinery (AREA)
  • Quality & Reliability (AREA)
  • General Factory Administration (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Testing And Monitoring For Control Systems (AREA)
US12/170,542 2008-07-10 2008-07-10 Methods and systems for pervasive diagnostics Active 2029-02-27 US7937175B2 (en)

Priority Applications (3)

Application Number Priority Date Filing Date Title
US12/170,542 US7937175B2 (en) 2008-07-10 2008-07-10 Methods and systems for pervasive diagnostics
EP09164975.6A EP2144129B1 (en) 2008-07-10 2009-07-09 Methods and systems for pervasive diagnostics
JP2009163423A JP5107974B2 (ja) 2008-07-10 2009-07-10 パーベイシブ診断法のための方法およびシステム

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
US12/170,542 US7937175B2 (en) 2008-07-10 2008-07-10 Methods and systems for pervasive diagnostics

Publications (2)

Publication Number Publication Date
US20100010654A1 US20100010654A1 (en) 2010-01-14
US7937175B2 true US7937175B2 (en) 2011-05-03

Family

ID=41172283

Family Applications (1)

Application Number Title Priority Date Filing Date
US12/170,542 Active 2029-02-27 US7937175B2 (en) 2008-07-10 2008-07-10 Methods and systems for pervasive diagnostics

Country Status (3)

Country Link
US (1) US7937175B2 (ja)
EP (1) EP2144129B1 (ja)
JP (1) JP5107974B2 (ja)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080294582A1 (en) * 2007-05-24 2008-11-27 Palo Alto Research Center Incorporated Dynamic domain abstraction through meta-analysis
US20100011255A1 (en) * 2008-07-10 2010-01-14 Palo Alto Research Center Incorporated Methods and systems for continously estimating persistent and intermittent failure probabilities for production resources
US20100010657A1 (en) * 2008-07-10 2010-01-14 Palo Alto Research Center Incorporated Methods and systems for active diagnosis through logic-based planning
US20100010845A1 (en) * 2008-07-10 2010-01-14 Palo Alto Research Center Incorporated Methods and systems for constructing production plans
US20100010952A1 (en) * 2008-07-10 2010-01-14 Palo Alto Research Center Incorporated Methods and systems for target value path identification
US20100241251A1 (en) * 2009-03-23 2010-09-23 Palo Alto Research Center Incorporated Methods and systems for fault diagnosis in observation rich systems
US20110172800A1 (en) * 2010-01-12 2011-07-14 Koizumi Ryuya Scheduler, substrate processing apparatus, and method of transferring substrates in substrate processing apparatus
US20120101863A1 (en) * 2010-10-22 2012-04-26 Byron Edwin Truax Machine-management system
US20140229409A1 (en) * 2008-03-08 2014-08-14 Tokyo Electron Limited Method and apparatus for self-learning and self-improving a semiconductor manufacturing tool
US9275335B2 (en) 2008-03-08 2016-03-01 Tokyo Electron Limited Autonomous biologically based learning tool
US9524570B1 (en) 2013-09-23 2016-12-20 The Boeing Company Visualizing real time production status

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2010211364A (ja) * 2009-03-09 2010-09-24 Hitachi Ltd プロジェクトのシミュレーション方法及びシミュレーションシステム
US8265777B2 (en) * 2009-06-02 2012-09-11 Palo Alto Research Center Incorporated Pervasive model adaptation: the integration of planning and information gathering in dynamic production systems
US8392455B2 (en) * 2009-07-02 2013-03-05 Palo Alto Research Center Incorporated Multi-interval heuristics for accelerating target-value search
US8364629B2 (en) 2009-07-02 2013-01-29 Palo Alto Research Center Incorporated Depth-first search for target value problems
DE102013220300A1 (de) * 2013-06-12 2014-12-18 Siemens Aktiengesellschaft Steuerung einer Anlage mit gleichartigen Anlagenteilen
US11501388B2 (en) * 2016-10-31 2022-11-15 Nec Corporation Production management apparatus, method, and non-transitory medium
US11580267B2 (en) * 2020-02-18 2023-02-14 Palo Alto Research Center Incorporated System and method for augmenting lumped parameter model diagnosis engine

Citations (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5369570A (en) * 1991-11-14 1994-11-29 Parad; Harvey A. Method and system for continuous integrated resource management
US5533413A (en) 1994-06-30 1996-07-09 Yokogawa Electric Corporation Equipment diagnosis system
US5914875A (en) 1996-01-11 1999-06-22 Kabushiki Kaisha Toshiba Method and apparatus for diagnosing plant anomaly
US6012152A (en) 1996-11-27 2000-01-04 Telefonaktiebolaget Lm Ericsson (Publ) Software fault management system
US6298308B1 (en) * 1999-05-20 2001-10-02 Reid Asset Management Company Diagnostic network with automated proactive local experts
US6326758B1 (en) 1999-12-15 2001-12-04 Reliance Electric Technologies, Llc Integrated diagnostics and control systems
US20040088207A1 (en) 2002-10-30 2004-05-06 Xerox Corporation Planning and scheduling reconfigurable systems around off-line resources
US20040085561A1 (en) * 2002-10-30 2004-05-06 Xerox Corporation Planning and scheduling reconfigurable systems with regular and diagnostic jobs
US6764267B2 (en) 1998-02-20 2004-07-20 F.R. Drake Patty loader and method
US20040267395A1 (en) * 2001-08-10 2004-12-30 Discenzo Frederick M. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US6847854B2 (en) * 2001-08-10 2005-01-25 Rockwell Automation Technologies, Inc. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US6898475B1 (en) 2004-05-27 2005-05-24 Palo Alto Research Center Inc. System and method utilizing temporal constraints to coordinate multiple planning sessions
US6999829B2 (en) * 2001-12-26 2006-02-14 Abb Inc. Real time asset optimization
US20060064291A1 (en) 2004-04-21 2006-03-23 Pattipatti Krishna R Intelligent model-based diagnostics for system monitoring, diagnosis and maintenance
US7043321B2 (en) 2004-05-27 2006-05-09 Palo Alto Research Center Incorporated Exception handling in manufacturing systems combining on-line planning and predetermined rules
US20060230313A1 (en) 2005-04-08 2006-10-12 Caterpillar Inc. Diagnostic and prognostic method and system
US7139629B2 (en) 2003-04-28 2006-11-21 Palo Alto Research Center Incorporated Planning and scheduling for failure recovery system and method
US7230736B2 (en) 2002-10-30 2007-06-12 Palo Alto Research Center, Incorporated Planning and scheduling reconfigurable systems with alternative capabilities
DE102006045804A1 (de) 2006-09-26 2008-03-27 Abb Patent Gmbh System und Verfahren zur optimierten Instandhaltungsplanung in einem Anlagen-überwachungssystem einer technischen Anlage oder eines technischen Prozesses
US20080148257A1 (en) 2006-12-19 2008-06-19 Palo Alto Research Center Incorporated Exception handling
US7558638B2 (en) * 2006-02-22 2009-07-07 Gm Global Technology Operations, Inc. Applying real-time control to a production system
US20090204237A1 (en) * 2001-08-10 2009-08-13 Rockwell Automation Technologies, Inc. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US20090210081A1 (en) * 2001-08-10 2009-08-20 Rockwell Automation Technologies, Inc. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US7627388B2 (en) * 2005-10-28 2009-12-01 Core, Inc. Reliability tools for complex systems

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3188812B2 (ja) * 1994-06-30 2001-07-16 横河電機株式会社 設備診断システム
JPH09190219A (ja) * 1996-01-11 1997-07-22 Toshiba Corp プラント異常診断装置

Patent Citations (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5369570A (en) * 1991-11-14 1994-11-29 Parad; Harvey A. Method and system for continuous integrated resource management
US5533413A (en) 1994-06-30 1996-07-09 Yokogawa Electric Corporation Equipment diagnosis system
US5914875A (en) 1996-01-11 1999-06-22 Kabushiki Kaisha Toshiba Method and apparatus for diagnosing plant anomaly
US6012152A (en) 1996-11-27 2000-01-04 Telefonaktiebolaget Lm Ericsson (Publ) Software fault management system
US6764267B2 (en) 1998-02-20 2004-07-20 F.R. Drake Patty loader and method
US6298308B1 (en) * 1999-05-20 2001-10-02 Reid Asset Management Company Diagnostic network with automated proactive local experts
US6326758B1 (en) 1999-12-15 2001-12-04 Reliance Electric Technologies, Llc Integrated diagnostics and control systems
US20090210081A1 (en) * 2001-08-10 2009-08-20 Rockwell Automation Technologies, Inc. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US20040267395A1 (en) * 2001-08-10 2004-12-30 Discenzo Frederick M. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US6847854B2 (en) * 2001-08-10 2005-01-25 Rockwell Automation Technologies, Inc. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US20090204237A1 (en) * 2001-08-10 2009-08-13 Rockwell Automation Technologies, Inc. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US6999829B2 (en) * 2001-12-26 2006-02-14 Abb Inc. Real time asset optimization
US20040088207A1 (en) 2002-10-30 2004-05-06 Xerox Corporation Planning and scheduling reconfigurable systems around off-line resources
US7230736B2 (en) 2002-10-30 2007-06-12 Palo Alto Research Center, Incorporated Planning and scheduling reconfigurable systems with alternative capabilities
US7233405B2 (en) * 2002-10-30 2007-06-19 Palo Alto Research Center, Incorporated Planning and scheduling reconfigurable systems with regular and diagnostic jobs
US20040085561A1 (en) * 2002-10-30 2004-05-06 Xerox Corporation Planning and scheduling reconfigurable systems with regular and diagnostic jobs
US7139629B2 (en) 2003-04-28 2006-11-21 Palo Alto Research Center Incorporated Planning and scheduling for failure recovery system and method
US20060064291A1 (en) 2004-04-21 2006-03-23 Pattipatti Krishna R Intelligent model-based diagnostics for system monitoring, diagnosis and maintenance
US7043321B2 (en) 2004-05-27 2006-05-09 Palo Alto Research Center Incorporated Exception handling in manufacturing systems combining on-line planning and predetermined rules
US6898475B1 (en) 2004-05-27 2005-05-24 Palo Alto Research Center Inc. System and method utilizing temporal constraints to coordinate multiple planning sessions
US20060230313A1 (en) 2005-04-08 2006-10-12 Caterpillar Inc. Diagnostic and prognostic method and system
US7627388B2 (en) * 2005-10-28 2009-12-01 Core, Inc. Reliability tools for complex systems
US7558638B2 (en) * 2006-02-22 2009-07-07 Gm Global Technology Operations, Inc. Applying real-time control to a production system
DE102006045804A1 (de) 2006-09-26 2008-03-27 Abb Patent Gmbh System und Verfahren zur optimierten Instandhaltungsplanung in einem Anlagen-überwachungssystem einer technischen Anlage oder eines technischen Prozesses
US20080148257A1 (en) 2006-12-19 2008-06-19 Palo Alto Research Center Incorporated Exception handling

Non-Patent Citations (9)

* Cited by examiner, † Cited by third party
Title
European Search Report dated Oct. 21, 2010.
European Search Report, EP 10 15 6820, Jul. 8, 2010.
Markus P.J. Fromherz, Daniel G. Bobrow, and Johan De Kleer, "Model-based Computing for Design and Control of Reconfigurable Systems", Palo Alto Research Center, Published in Al Magazine, Special Issue on Qualitative Reasoning, vol. 24, n. 4, 2003 pp. 120-130.
Minh Do and Wheeler Ruml, "Lessons Learned in Applying Domain-Independent Planning to High-Speed Manufacturing", Palo Alto Research Center, 2006.
Patrick Haslum and Hector Geffner, "Heuristic Planning with Time and Resources", Department of Computer Science, Linkoping University, Sweden, Departamento de Computacion, Universidad Simon Bolivar, Venezuela.
Peter E. Hart, Nils J. Nilsson, and Bertram Raphael, "A Formal Basis for the Heuristic Determination of Minimum Cost Paths", IEEE Transactions of Systems Science and Cybernetics, vol. ssc-4, No. 2, Jul. 1968.
Richard Dearden and Dan Clancy, "Particle Filters for Real-Time Fault Detection in Planetary Rovers", Research Institute for Advanced Computer Science, NASA Ames Research Center.
Rina Dechter and Judea Pearl, "Generalized Best-First Search Strategies and the Optimality of A", University of California, Los Angeles, California, Journal of the Association for Computing Machinery, vol. 32, No. 3, Jul. 1985, pp. 505-536.
Roberto Cervoni, Amedeo Cesta, and Angelo Oddi, "Managing Dynamic Temporal Constraint Networks", National Research Council of Italy.

Cited By (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8073798B2 (en) * 2007-05-24 2011-12-06 Palo Alto Research Center Incorporated Dynamic domain abstraction through meta-analysis
US20080294582A1 (en) * 2007-05-24 2008-11-27 Palo Alto Research Center Incorporated Dynamic domain abstraction through meta-analysis
US9424528B2 (en) * 2008-03-08 2016-08-23 Tokyo Electron Limited Method and apparatus for self-learning and self-improving a semiconductor manufacturing tool
US9275335B2 (en) 2008-03-08 2016-03-01 Tokyo Electron Limited Autonomous biologically based learning tool
US20140229409A1 (en) * 2008-03-08 2014-08-14 Tokyo Electron Limited Method and apparatus for self-learning and self-improving a semiconductor manufacturing tool
US8145334B2 (en) * 2008-07-10 2012-03-27 Palo Alto Research Center Incorporated Methods and systems for active diagnosis through logic-based planning
US8266092B2 (en) 2008-07-10 2012-09-11 Palo Alto Research Center Incorporated Methods and systems for target value path identification
US20100011255A1 (en) * 2008-07-10 2010-01-14 Palo Alto Research Center Incorporated Methods and systems for continously estimating persistent and intermittent failure probabilities for production resources
US20100010952A1 (en) * 2008-07-10 2010-01-14 Palo Alto Research Center Incorporated Methods and systems for target value path identification
US8165705B2 (en) 2008-07-10 2012-04-24 Palo Alto Research Center Incorporated Methods and systems for continuously estimating persistent and intermittent failure probabilities for production resources
US20100010657A1 (en) * 2008-07-10 2010-01-14 Palo Alto Research Center Incorporated Methods and systems for active diagnosis through logic-based planning
US8219437B2 (en) 2008-07-10 2012-07-10 Palo Alto Research Center Incorporated Methods and systems for constructing production plans
US20100010845A1 (en) * 2008-07-10 2010-01-14 Palo Alto Research Center Incorporated Methods and systems for constructing production plans
US8359110B2 (en) 2009-03-23 2013-01-22 Kuhn Lukas D Methods and systems for fault diagnosis in observation rich systems
US20100241251A1 (en) * 2009-03-23 2010-09-23 Palo Alto Research Center Incorporated Methods and systems for fault diagnosis in observation rich systems
US8655472B2 (en) * 2010-01-12 2014-02-18 Ebara Corporation Scheduler, substrate processing apparatus, and method of transferring substrates in substrate processing apparatus
US20110172800A1 (en) * 2010-01-12 2011-07-14 Koizumi Ryuya Scheduler, substrate processing apparatus, and method of transferring substrates in substrate processing apparatus
US20120101863A1 (en) * 2010-10-22 2012-04-26 Byron Edwin Truax Machine-management system
US9524570B1 (en) 2013-09-23 2016-12-20 The Boeing Company Visualizing real time production status

Also Published As

Publication number Publication date
JP5107974B2 (ja) 2012-12-26
US20100010654A1 (en) 2010-01-14
EP2144129A2 (en) 2010-01-13
EP2144129B1 (en) 2013-06-19
EP2144129A3 (en) 2010-12-01
JP2010020772A (ja) 2010-01-28

Similar Documents

Publication Publication Date Title
US7937175B2 (en) Methods and systems for pervasive diagnostics
US8219437B2 (en) Methods and systems for constructing production plans
US8145334B2 (en) Methods and systems for active diagnosis through logic-based planning
US8266092B2 (en) Methods and systems for target value path identification
US8165705B2 (en) Methods and systems for continuously estimating persistent and intermittent failure probabilities for production resources
US6856845B2 (en) Monitoring and reporting incremental job status system and method
US8359110B2 (en) Methods and systems for fault diagnosis in observation rich systems
JP5542697B2 (ja) 生物学に基づく自律学習ツール
US7139629B2 (en) Planning and scheduling for failure recovery system and method
JP5330316B2 (ja) パーベイシブ・モデル適応：動的生産システムにおける計画と情報収集の統合
EP2000899A2 (en) System and method for real-time system control using precomputed plans
JP2008256981A (ja) 故障診断システム
Miyashita Case-based knowledge acquisition for schedule optimization
Sharp et al. Selecting Optimal Data for Creating Informed Maintenance Decisions in a Manufacturing Environment
Kuhn et al. Pervasive diagnosis
KR20240102488A (ko) 생산-에너지 데이터 특성에 따른 공정설비 최적운영 시스템
Cai Semiconductor manufacturing inspired integrated scheduling problems: production planning, advanced process control, and predictive maintenance
Zeng Condition monitoring and supply chain coordination for spare parts inventory management
JPWO2023188214A5 (ja)

Legal Events

Date Code Title Description
AS Assignment

Owner name: PALO ALTO RESEARCH CENTER INCORPORATED, CALIFORNIA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:DE KLEER, JOHAN;PRICE, ROBERT;KUHN, LUKAS D.;AND OTHERS;REEL/FRAME:021361/0392

Effective date: 20080728

FEPP Fee payment procedure

Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

STCF Information on status: patent grant

Free format text: PATENTED CASE

FPAY Fee payment

Year of fee payment: 4

MAFP Maintenance fee payment

Free format text: PAYMENT OF MAINTENANCE FEE, 8TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1552); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

Year of fee payment: 8

MAFP Maintenance fee payment

Free format text: PAYMENT OF MAINTENANCE FEE, 12TH YEAR, LARGE ENTITY (ORIGINAL EVENT CODE: M1553); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

Year of fee payment: 12

AS Assignment

Owner name: XEROX CORPORATION, CONNECTICUT

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:PALO ALTO RESEARCH CENTER INCORPORATED;REEL/FRAME:064038/0001

Effective date: 20230416

AS Assignment

Owner name: CITIBANK, N.A., AS COLLATERAL AGENT, NEW YORK

Free format text: SECURITY INTEREST;ASSIGNOR:XEROX CORPORATION;REEL/FRAME:064760/0389

Effective date: 20230621

AS Assignment

Owner name: XEROX CORPORATION, CONNECTICUT

Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE REMOVAL OF US PATENTS 9356603, 10026651, 10626048 AND INCLUSION OF US PATENT 7167871 PREVIOUSLY RECORDED ON REEL 064038 FRAME 0001. ASSIGNOR(S) HEREBY CONFIRMS THE ASSIGNMENT;ASSIGNOR:PALO ALTO RESEARCH CENTER INCORPORATED;REEL/FRAME:064161/0001

Effective date: 20230416

AS Assignment

Owner name: JEFFERIES FINANCE LLC, AS COLLATERAL AGENT, NEW YORK

Free format text: SECURITY INTEREST;ASSIGNOR:XEROX CORPORATION;REEL/FRAME:065628/0019

Effective date: 20231117

AS Assignment

Owner name: XEROX CORPORATION, CONNECTICUT

Free format text: TERMINATION AND RELEASE OF SECURITY INTEREST IN PATENTS RECORDED AT RF 064760/0389;ASSIGNOR:CITIBANK, N.A., AS COLLATERAL AGENT;REEL/FRAME:068261/0001

Effective date: 20240206

Owner name: CITIBANK, N.A., AS COLLATERAL AGENT, NEW YORK

Free format text: SECURITY INTEREST;ASSIGNOR:XEROX CORPORATION;REEL/FRAME:066741/0001

Effective date: 20240206